On Key Techniques For Random Access Systems In IoT-Oriented Satellite Networks

The satellite Internet-of-Things(IoT)is one of the important application scenarios in the field of wireless communications,and it is also an important part of the B5 G and airspace-ground integrated networks.Random access(RA)has become a typical multiple access method in the satellite IoT because it does not require resource allocation and central scheduling,and it has also received extensive attention.However,with the increase in the number of satellite IoT devices and the intelligent and energy-efficient development of satellite IoT,RA is also constantly facing new challenges.On the one hand,the massive number of satellite IoT devices puts huge pressure on the RA of the satellite IoT,which will lead to severe collision between accessing devices and even network congestion;On the other hand,the satellite IoT devices are gradually miniaturized and intelligentized,which poses huge challenges to device's computational complexity,transmission power,and energy consumption.In order to cope with these issues,research on key techniques of the RA for satellite IoT has been carried out from three scenarios: RA for inter-satellite non-cooperative satellites,RA for inter-satellite cooperative satellites,and RA for relay-assisted satellites,and many effective satellite RA protocols are proposed in terms of different metrics.The main contributions of the thesis are listed as follows:1.In terms of the RA for inter-satellite non-cooperative satellites,in order to meet the highthroughput demand of the future satellite IoT and address the problems of low throughput of the existing satellite RA protocols,the research on high-throughput RA for inter-satellite non-cooperative satellites has been carried out.The pre-weighting based contention resolution diversity slotted ALOHA(PW-CRDSA)protocol is first proposed,which adopts the proposed joint multi-slot multi-user detection(JMMD)algorithm to resolve packet collision.The throughputs of the RA systems are thus improved.The random pattern multiplexing based CRDSA(RPM-CRDSA)protocol is then proposed,in which resource multiplexing is used to perform the RA for users.At the same time,the message passing detection(MPD)algorithm is used to achieve multi-user detection.The joint design of the transceiver makes the throughputs of the RA systems are greatly improved.Finally,the spatial group based optimal uplink power control scheme is proposed.The scheme first divides users into multiple groups and performs uplink power control in a distributed manner,and then the number of the groups and target received power levels are jointly optimized to maximize the throughputs of the RA systems.The throughputs of the three proposed RA protocols are greatly improved compared with the conventional RA protocols.2.In terms of the RA for inter-satellite non-cooperative satellites,in order to address the energy-limited problems of the satellite IoT devices,the research on the energy-limited RA for inter-satellite non-cooperative satellites has been carried out.The energy-limited optimal CRDSA protocol is first proposed.This protocol maximizes the achievable sum rate(ASR)of the system by jointly optimizing the packet transmit power and the detection threshold under the total transmit power conditions for each user.The CRDSA protocol is extended to IRSA protocol,and the energy-limited optimal irregular repetition slotted ALOHA(IRSA)protocol is then proposed.Under the total transmit power,this protocol causes transmit power diversity by sending different numbers of packet replicas,and a semi-analytical(SEA)model is used to approximate the packet error rate(PER)of the different code rates.The optimal user degree distributions under different code rates and total transmit powers conditions are obtained using differential evolution algorithm,which maximize the system throughput.In the energy-limited scenarios,the two proposed RA protocols outperform the conventional RA protocols in terms of the system throughput.3.In terms of the RA for inter-satellite cooperative satellites,in order to address the lowthroughput and inter-satellite interference problems of the RA for inter-satellite non-cooperative satellites,the research on the RA for inter-satellite cooperative satellites has been carried out.First,considering the complete overlap coverage area between multiple satellites,the multi-satellite cooperative CRDSA(MSC-CRDSA)protocol is proposed,which realizes the decoding of collided packets through multi-satellite joint detection,and further improves the throughput by combining the interference cancellation technology.Then,considering that there is only a partial overlap coverage area between multiple satellites and transmit power diversity is introduced,the optimal MSC-CRDSA protocol based on power diversity is proposed.This protocol combines multi-satellite cooperative detection and transmit power diversity,which greatly improves the multi-satellite cooperative throughput.Besides,the throughput of the MSC-CRDSA protocol is maximized by optimizing the probability distribution of transmit power.The RA for inter-satellite cooperative satellites can lead to larger throughput improvement with respect to the RA for inter-satellite non-cooperative satellites.4.In terms of the RA for relay-assisted satellites,in order to address the problem that some satellite IoT devices can not directly access the satellites due to the transmit power limitation or the block of line of sight transmission,a new relay-assisted satellite RA architecture is proposed and the research on the RA for relay-assisted satellites has been carried out.First,we consider satellite users as relays,and this type of relays need to forward signals from non-relay users in addition to its own signals.For this reason,an optimal user pairing and power allocation scheme is proposed.An optimal user pairing problem is formulated and solved by the proposed Q-learning based distributed user pairing algorithm(QL-DUPA),and the power allocation between each user pair is optimized using convex optimization method.Finally,the proposed optimal user pairing and power allocation scheme is combined with RA protocols,which greatly improves the sum rates of RA systems.Then we consider that the relay itself does not need to send the signal but only forwards the non-relay user's signal to the satellite.In this scenario,distributed Q learning based joint relay selection and access control(DQL-JRSAC)scheme is proposed for the satellite terrestrial relay networks.This scheme improves the learning efficiency of Q-learning through a semi-random access(SRA)architecture,and constructs joint relay selection and access control(JRSAC)problem.Finally,the use of the distributed Q-learning(DQL)achieves the optimal relay selection and access control.